Cartographie avec R

Cartographie avec R : le package cartography

SEMIN-R
Museum National d’Histoire Naturelle | Vendredi 16 juin 2017

Timothée Giraud
UMS RIATE
http://rgeomatic.hypotheses.org/

Les fonctionnalités spatiales de R

Les indispensables

Import / Export

rgdal est une interface entre R et les librairies GDAL (Geospatial Data Abstraction Library) et PROJ4.

library("rgdal")
nuts3 <- readOGR(dsn = "data", layer = "nuts3", verbose = TRUE)

OGR data source with driver: ESRI Shapefile 
Source: "data", layer: "nuts3"
with 1448 features
It has 7 fields

Manipulation et affichage

sp fournit des classes et des methodes pour les données spatiales dans R.

library("sp")
head(nuts3@data)

	id	birth_2008	death_2008	gdppps1999	gdppps2008	pop1999	pop2008
0	AT111	272	445	509	641	39148.91	37452
1	AT112	1195	1480	2262	3272	137469.46	146383
2	AT113	748	1142	1368	1783	100113.63	97350
3	AT121	2480	2119	4091	5807	238591.26	241083
4	AT122	2266	2577	4130	5774	244959.00	251838
5	AT123	1363	1392	3098	4504	144269.11	147468

plot(nuts3)

plot(nuts3, col = "#DAE3E6", border = "#8A0641", lwd = 0.5)

Géotraitements

rgeos donne accès à la librairie d’opérations spatiales GEOS (Geometry Engine - Open Source) qui permet notamment d’effectuer les géotraitements suivants :

• Area / Perimeter
• Distances
  
• Dissolve
• Buffer
• Overlap / intersect / difference
• Contains / within
• Union
• …

Agrégation des polygones / dissolve

library("rgeos")
europe <- gUnaryUnion(spgeom = nuts3)
plot(nuts3, lwd = 0.5)
plot(europe, lwd = 2, border = "red", add=T)

Création de zones tampons / buffer

library("rgeos")
europeBuffer <- gBuffer(spgeom = europe, width = 50000)
plot(europe, col = "#92C5D6")
plot(europeBuffer, add = T, border = "red")

Le futur : le package sf

Première release : 31 octobre 2016
Auteur principal et maintainer : Edzer Pebesma
Financement :

Import / Export

Manipulation et affichage

Géotraitements

Le package cartography

Installation

• Version stable (CRAN)

install.packages("cartography")

• Version de développement (Github)

Cette version permet d’utiliser les objets sf.

devtools::install_github(repo = "Groupe-ElementR/cartography", ref = "devsf")

Utilisation

Cartes choroplèthes

library(cartography)

# chargement de données
data(nuts2006)

# Calcul du taux de croissance annuel moyen
nuts2.df$cagr <- 100 * (((nuts2.df$pop2008 / nuts2.df$pop1999)^(1/9)) - 1) 

# Cartographie
choroLayer(spdf = nuts2.spdf, df = nuts2.df, var = "cagr")
title("Taux de croissance en Europe")

Après ce premier jet, il est ensuite possible de paramétrer très finement la carte : palette de couleurs, discrétisation, légende, couches d’habillage…

# Construire une palette de couleurs
cols <- carto.pal(pal1 = "green.pal", n1 = 2, 
                  pal2 = "red.pal", n2 = 4) 

# Affichage de couches d'habillage
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
plot(world.spdf, col  = "#E3DEBF", border=NA, add=TRUE)

# Cartographie du taux de croissance annuel moyen
choroLayer(spdf = nuts2.spdf, df = nuts2.df, var = "cagr",
           breaks = c(-2.43,-1.0,0.0,0.5,1.0,2.0,3.1), 
           col = cols,
           border = "grey40",
           lwd = 0.5, 
           legend.pos = "right",
           legend.title.txt = "taux de croissance\nannuel moyen", 
           legend.values.rnd = 2, 
           add = TRUE) 

# Affichage de couches d'habillage
plot(nuts0.spdf,border = "grey20", lwd=0.75, add=TRUE)

# Ajout des titres, légende, sources, etc.
layoutLayer(title = "Taux de croissance en Europe", 
            author = "cartography", 
            sources = "Eurostat, 2008", frame = TRUE, col = NA, 
            scale = NULL,coltitle = "black",
            south = TRUE) 

Cartes en symboles proportionnels

Cartographie d’un stock (la population nationale) avec des figurés proportionnels.

# Affichage de couches d'habillage
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
plot(world.spdf, col  = "#E3DEBF", border=NA, add=TRUE)
plot(nuts0.spdf, col = "#D1914D",border = "grey80", add=TRUE)

# Cartographie de la population des pays en cercles proportionnels
propSymbolsLayer(spdf = nuts0.spdf, df = nuts0.df,
                 var = "pop2008", 
                 symbols = "circle", col =  "seagreen4",
                 legend.pos = "right", inches = 0.35,
                 legend.title.txt = "Total\npopulation (2008)",
                 legend.style = "c")

# Ajout de labels
dflab <- nuts0.df[order(nuts0.df$pop2008, decreasing = TRUE),][1:8,]
dflab$lab <- paste(dflab$id, "\n", round(dflab$pop2008/1000000,0), "M", sep ="")

# Label plot of the 8 most populated countries
labelLayer(spdf = nuts0.spdf, 
           df = dflab, 
           txt = "lab", 
           col = "#690409", 
           cex = 0.8, 
           font = 2) 

# Ajout des titres, légende, sources, etc.
layoutLayer(title = "Countries Population in Europe",
            theme = 'green.pal',
            frame = FALSE,
            author = "cartography",
            sources = "Eurostat, 2008",
            scale = NULL,
            south = TRUE)

Cartes en symboles proportionnels colorés

# Load data
data(nuts2006)

# Compute the compound annual growth rate
nuts2.df$cagr <- (((nuts2.df$pop2008 / nuts2.df$pop1999)^(1/9)) - 1) * 100

# Plot a layer with the extent of the EU28 countries with only a background color
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
# Plot non european space
plot(world.spdf, col  = "#E3DEBF", border = NA, add = TRUE)
# Plot Nuts2 regions
plot(nuts2.spdf, col = "grey60",border = "white", lwd = 0.4, add = TRUE)

# Set a custom color palette
cols <- carto.pal(pal1 = "blue.pal", n1 = 2, pal2 = "red.pal", n2 = 4)

# Plot symbols with choropleth coloration
propSymbolsChoroLayer(spdf = nuts2.spdf, 
                      df = nuts2.df, 
                      var = "pop2008", 
                      inches = 0.1, 
                      var2 = "cagr", 
                      col = cols, 
                      breaks = c(-2.43,-1,0,0.5,1,2,3.1), 
                      border = "grey50",  
                      lwd = 0.75, 
                      legend.var.pos = "topright", 
                      legend.var.values.rnd = -3,
                      legend.var.title.txt = "Total Population", 
                      legend.var.style = "e", 
                      legend.var2.pos = "right", 
                      legend.var2.title.txt = "Compound Annual\nGrowth Rate") 

# layout
layoutLayer(title = "Demographic trends, 1999-2008", coltitle = "black",
            sources = "Eurostat, 2011", scale = NULL,
            author = "cartography", frame ="", col = NA)

Cartes de flux

Il s’agit de représenter des données, agrégées à un niveau régional, sur les jumelages entre villes.

# Données sur les jumelages
head(twincities.df)

i	j	fij
DE14	AT11	1
DE21	AT11	1
DE23	AT11	1
DE26	AT11	2
DE91	AT11	1
DEB3	AT11	1

# Creation d'une couche de liens
twincities.sf <- getLinkLayer(x = nuts2.spdf, df = twincities.df[,1:2])

# Affichage des liens créés
plot(st_geometry(twincities.sf), lwd = 0.2)

# Affichage de couches d'habillage
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
plot(world.spdf, col  = "#E3DEBF", border=NA, add=TRUE)
plot(nuts2.spdf, col = "#D1914D",border = "grey80", add=TRUE)

# Cartographie des liens
gradLinkLayer(x = twincities.sf, df = twincities.df,   
              var = "fij", 
              breaks = c(2,5,15,20,30), 
              lwd = c(0.1,1,4,10), 
              col = "#92000090",
              legend.pos = "right", legend.frame = TRUE,
              legend.title.txt = "Number of Agreements\n(regional level)",
              add = TRUE)

# Ajout des titres, légende, sources, etc.
layoutLayer(title = "International Twinning Agreements Between Cities", 
            author = "cartography", 
            sources = "Sources: Adam Ploszaj & Wikipedia, 2011",
            scale = NULL, south = TRUE, frame = TRUE, col = NA, 
            coltitle = "black")

Discontinuités

# Load data
data(nuts2006)

# Get a SpatialLinesDataFrame of countries borders
nuts0.contig <- getBorders(spdf = nuts0.spdf)

plot(nuts0.spdf, col = "grey", border = NA)
plot(st_geometry(nuts0.contig), 
     col = 1:nrow(nuts0.contig), 
     add=T, lwd = 4)

# Get the GDP per capita
nuts0.df$gdpcap <- nuts0.df$gdppps2008/nuts0.df$pop2008*1000000

# Plot a layer with the extent of the EU28 countries with only a background color
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
# Plot non european space
plot(world.spdf, col  = "#E3DEBF", border=NA, add=TRUE)

# Plot GDP per capita with a choropleth layer
choroLayer(spdf = nuts0.spdf, df = nuts0.df, var = "gdpcap", border = "grey80",
           col = carto.pal(pal1 = "kaki.pal", n1 = 6), method = "quantile",
           nclass = 6, add=TRUE, legend.pos = "right", 
           legend.values.rnd = -2,
           legend.title.txt = "GDP per Capita\n(in euros)")

# Plot discontinuities
discLayer(x = nuts0.contig, # sf of borders
          df = nuts0.df, # data frame on countries
          var = "gdpcap", # variable used to compute discontinuties 
          type = "rel", # type of discontinuity measure 
          method="equal", # discretisation of discontinuities
          nclass=4, # number of discontinuities classes
          threshold = 0.5, # representation threshold of discontinuities  
          sizemin = 0.5, # minimum size of discontinuities lines
          sizemax = 6, # maximum size of discontinuities lines
          col="red", # color of the lines
          legend.values.rnd = 1,
          legend.title.txt = "Discontinuities in \nGDP per Capita\n(relative)",
          legend.pos = "topright", 
          add=TRUE)

# Layout
layoutLayer(title = "Wealth Disparities in Europe", coltitle = "black",
            sources = "Eurostat, 2011", scale = NULL,
            author = "cartography", frame ="", col = NA)

Carroyages

library(cartography)
# Load data
data(nuts2006)

# Create a grid layer
nuts2.spdf@data <- nuts2.df
mygrid <- getGridLayer(x = nuts2.spdf, 
                       cellsize = 200000 * 200000, 
                       var = "pop2008")
                       
# Plot dentsity of population
## conversion from square meter to square kilometers
mygrid$densitykm <- mygrid$pop2008 * 1000 * 1000 / mygrid$gridarea

# Plot a layer with the extent of the EU28 countries with only a background color
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")
# Plot non european space
plot(world.spdf, col  = "#E3DEBF", border=NA, add=TRUE)

# Set a custom color palette
cols <- carto.pal(pal1 = "wine.pal", n1 = 6)

# Plot the gridded population density
choroLayer(x = mygrid, var = "densitykm", 
           border = "grey80", col = cols, legend.pos = "topright",
           method = "q6", add = TRUE, legend.values.rnd = 1,
           legend.title.txt = "Population Density\n(inhabitant/km²)")

# Layout
layoutLayer(title = "Population Density", coltitle = "black",
            sources = "Eurostat, 2011", scale = NULL, 
            author = "cartography", frame ="", col = NA)

Cartes lissées

# set margins
opar <- par(mar = c(0, 0, 1.2, 0))
# Load data
data(nuts2006)

nuts3.spdf@data = nuts3.df
# Create a grid layer
mygrid <- getGridLayer(x = sf::st_as_sf(nuts3.spdf), 
                       cellsize = 50000 * 50000, 
                       type = "regular", 
                       var = c("pop2008", "gdppps2008"))


# Compute data for the grid layer
mygrid$gdp <- mygrid$gdppps2008*1000000

# list of breaks
v <- c(2920, 5000, 10000, 15000, 20000, 23500, 30000, 35000, 40000, 42720)
# Plot a layer with the extent of the EU28 countries with only a background
# color
plot(nuts0.spdf, border = NA, col = NA, bg = "#A6CAE0")

# Plot non european space
plot(world.spdf, col = "#E3DEBF", border = NA, add = TRUE)

# set a color palette
cols <- c(rev(carto.pal("green.pal", 5)), carto.pal("orange.pal", 4))

# compute & display the potential map
smoothLayer(x = mygrid, var = "gdp", var2 = "pop2008", breaks = v, 
            span = 1e+05, beta = 2, mask = nuts0.spdf, resolution = 49000, col = cols, 
            legend.title.txt = "Potential\nGDP per capita\n(in euros)", legend.values.rnd = -2, 
            border = "grey80", lwd = 0.5, add = T, legend.pos = "topright")

# plot Europe contour
plot(rgeos::gBuffer(nuts0.spdf, FALSE, 1), add = T, col = NA, border = "grey50")

# plot a layout
layoutLayer(title = "Wealth Inequalities in Europe, 2008", 
            author = "Package cartography v2.0.0", 
            sources = "Source: Eurostat, 2011", frame = TRUE, scale = 500, north = FALSE, 
            theme = "sand.pal")

# plot a text on the map
text(x = 6271272, y = 3743765, labels = "Distance function:\n- type = exponential\n- beta = 2\n- span = 100 km", 
     cex = 0.8, adj = 0, font = 3)

Ressources

La présentation est accessible à cette adresse :
https://rcarto.github.io/semin-r_2017/index.html

Son code source est accessible sur GitHub.

Package cartography

La page GitHub du package cartography

C’est dans ce dépôt GitHub que se déroule le développement du package et que se tiennent les discussions à son sujet.

Billets de blogs, tutoriels, présentations

Demo codes in the R graph Gallery (EN)
Create and integrate maps in your R workflow with the cartography package (EN)
De superbes cartes thématiques…(FR)
Le package cartography a un an(FR)
Cartographie avec R : le package cartography (FR)
R pour les cartographes (FR)
Comment faire un carton avec R? (FR - How to build inset maps)
Tutoriel - Cartographie avec R (FR)
Cartographie et traitement de l’information géographique avec R (FR)
R pour les cartographes : le package cartography (FR)

Le blog http://rgeomatic.hypotheses.org/

Package sf

La page GitHub du package sf

C’est dans ce dépôt GitHub que se déroule le développement du package et que se tiennent les discussions à son sujet.

Les ressources publiées par Edzer Pebesma

Le créateur et maintainer de sf a publié un certain de nombre de vignettes et de billets autour du package.

• Vignettes

  • Simple Features for R Explication très détaillée des principes et du fonctionnement du package.
  • Reading, Writing and Converting Simple Features Comment importer et exporter des objets.
  • Manipulating Simple Feature Geometries Comment manipuler les objets sf, géotraitements, projections…

• Billets de blog

  • Simple features for R
  • Simple features for R, part 2
  • Simple features now on CRAN
  • sf - plot, graticule, transform, units, cast, is

• Migration de sp ver sf

  Un tableau de migration de sp vers sf pour passer de l’utilisation des packages sp, rgeos et rgdal à l’utilisation de sf.

Billets de blogs / tutoriels

• Faire des cartes avec R - Introduction au package sf - Santiago Mora

• Tidy spatial data in R: using dplyr, tidyr, and ggplot2 with sf - Matt Strimas-Mackey

• First Impressions From sf – The Simple Features R Package - Geographic Data Science Lab

• Spatial analysis pipelines with simple features in R - Kyle Walker

• Spatial analysis in R with the sf package - rhodyRstats

• Landscape Ecology with R:A tutorial with raster and sf - Jeffrey W. Hollister

• Mapping “France at night” with the new sf package - Sharp Sight Labs

La cartographie

Béguin & Pumain (2003)

• Michelle Béguin et Denise Pumain. “La représentation des données géographiques, Statistique et cartographie.” (2003). Paris, Armand Colin, Coll. Cursus, 192p.

!

• Nicolas Lambert et Christine Zanin. “Manuel de cartographie: principes, méthodes, applications.” (2016). Paris, Armand Colin, Coll. Cursus, 224p.